1 1  Capabilities: Serial (C), shared-memory (OpenMP or Pthreads), distributed-memory (hybrid MPI+ OpenM + CUDA). All have Fortran interface. Sparse LU.

Slides:



Advertisements
Similar presentations
EXTENDED MHD MODELING: BACKGROUND, STATUS AND VISION
Advertisements

CHAPTER 1: COMPUTATIONAL MODELLING
Point-wise Discretization Errors in Boundary Element Method for Elasticity Problem Bart F. Zalewski Case Western Reserve University Robert L. Mullen Case.
Lecture 8 Hybrid Solvers based on Domain Decomposition Xiaoye Sherry Li Lawrence Berkeley National Laboratory, USA crd-legacy.lbl.gov/~xiaoye/G2S3/
Numerical Parallel Algorithms for Large-Scale Nanoelectronics Simulations using NESSIE Eric Polizzi, Ahmed Sameh Department of Computer Sciences, Purdue.
Extending the capability of TOUGHREACT simulator using parallel computing Application to environmental problems.
Multilevel Incomplete Factorizations for Non-Linear FE problems in Geomechanics DMMMSA – University of Padova Department of Mathematical Methods and Models.
Two Approaches to Multiphysics Modeling Sun, Yongqi FAU Erlangen-Nürnberg.
1 Aug 7, 2004 GPU Req GPU Requirements for Large Scale Scientific Applications “Begin with the end in mind…” Dr. Mark Seager Asst DH for Advanced Technology.
PETSc Portable, Extensible Toolkit for Scientific computing.
6/22/2005ICS'20051 Parallel Sparse LU Factorization on Second-class Message Passing Platforms Kai Shen University of Rochester.
Direct and iterative sparse linear solvers applied to groundwater flow simulations Matrix Analysis and Applications October 2007.
1 Parallel Simulations of Underground Flow in Porous and Fractured Media H. Mustapha 1,2, A. Beaudoin 1, J. Erhel 1 and J.R. De Dreuzy IRISA – INRIA.
Lecture 2 Sparse matrix data structures, graphs, manipulation Xiaoye Sherry Li Lawrence Berkeley National Laboratory, USA crd-legacy.lbl.gov/~xiaoye/G2S3/
Presented by High-Performance Computing in Magnetic Fusion Energy Research Donald B. Batchelor RF Theory Plasma Theory Group Fusion Energy Division.
SuperLU: Sparse Direct Solver X. Sherry Li SIAM Short Course on the ACTS Collection Feburary 27, 2004.
MUMPS A Multifrontal Massively Parallel Solver IMPLEMENTATION Distributed multifrontal.
Iterative and direct linear solvers in fully implicit magnetic reconnection simulations with inexact Newton methods Xuefei (Rebecca) Yuan 1, Xiaoye S.
Ju LiKord SmithBen ForgetFelix ParaSid Yip Computational Nuclear Science and Engineering (New) H-LEVEL Grad Credit (Spring) Prerequisites:
Massively Parallel Magnetohydrodynamics on the Cray XT3 Joshua Breslau and Jin Chen Princeton Plasma Physics Laboratory Cray XT3 Technical Workshop Nashville,
Scalabilities Issues in Sparse Factorization and Triangular Solution Sherry Li Lawrence Berkeley National Laboratory Sparse Days, CERFACS, June 23-24,
SciDAC Software Introduction Osni Marques Lawrence Berkeley National Laboratory DOE Workshop for Industry Software Developers March 31,
S.S. Yang and J.K. Lee FEMLAB and its applications POSTEC H Plasma Application Modeling Lab. Oct. 25, 2005.
Overview of MHD and extended MHD simulations of fusion plasmas Guo-Yong Fu Princeton Plasma Physics Laboratory Princeton, New Jersey, USA Workshop on ITER.
Lecture 5 Parallel Sparse Factorization, Triangular Solution
Amesos Sparse Direct Solver Package Ken Stanley, Rob Hoekstra, Marzio Sala, Tim Davis, Mike Heroux Trilinos Users Group Albuquerque 3 Nov 2004.
PDE2D, A General-Purpose PDE Solver Granville Sewell Mathematics Dept. University of Texas El Paso.
Nonlinear Extended MHD Simulation Using High-Order Finite Elements C. Sovinec, C. Kim, H. Tian, Univ. of WI-Madison D. Schnack, A. Pankin, Science Apps.
1 1  Capabilities: Dynamic load balancing and static data partitioning -Geometric, graph-based, hypergraph-based -Interfaces to ParMETIS, PT-Scotch, PaToH.
1 Incorporating Iterative Refinement with Sparse Cholesky April 2007 Doron Pearl.
1 SciDAC TOPS PETSc Work SciDAC TOPS Developers Satish Balay Chris Buschelman Matt Knepley Barry Smith.
On the Use of Sparse Direct Solver in a Projection Method for Generalized Eigenvalue Problems Using Numerical Integration Takamitsu Watanabe and Yusaku.
Computational Aspects of Multi-scale Modeling Ahmed Sameh, Ananth Grama Computing Research Institute Purdue University.
1 1 What does Performance Across the Software Stack mean?  High level view: Providing performance for physics simulations meaningful to applications 
Amesos Interfaces to sparse direct solvers October 15, :30 – 9:30 a.m. Ken Stanley.
1 1  Capabilities: Scalable algebraic solvers for PDEs Freely available and supported research code Usable from C, C++, Fortran 77/90, Python, MATLAB.
Integrated Modeling for Burning Plasmas Workshop (W60) on “Burning Plasma Physics and Simulation 4-5 July 2005, University Campus, Tarragona, Spain Under.
1 1  Capabilities: Building blocks for block-structured AMR codes for solving time-dependent PDE’s Functionality for [1…6]D, mixed-dimension building.
Parallel and Distributed Computing Research at the Computing Research Institute Ananth Grama Computing Research Institute and Department of Computer Sciences.
PaStiX : how to reduce memory overhead ASTER meeting Bordeaux, Nov 12-14, 2007 PaStiX team LaBRI, UMR CNRS 5800, Université Bordeaux I Projet ScAlApplix,
Walter Schostak Center for Materials Under eXtreme Environment
Implementing Hypre- AMG in NIMROD via PETSc S. Vadlamani- Tech X S. Kruger- Tech X T. Manteuffel- CU APPM S. McCormick- CU APPM Funding: DE-FG02-07ER84730.
October 2008 Integrated Predictive Simulation System for Earthquake and Tsunami Disaster CREST/Japan Science and Technology Agency (JST)
Cracow Grid Workshop, November 5-6, 2001 Concepts for implementing adaptive finite element codes for grid computing Krzysztof Banaś, Joanna Płażek Cracow.
Department of Electronic Engineering, Tsinghua University Nano-scale Integrated Circuit and System Lab. Performance Analysis of Parallel Sparse LU Factorization.
1 A Proposal for a SWIM Slow-MHD 3D Coupled Calculation of the Sawtooth Cycle in the Presence of Energetic Particles Josh Breslau Guo-Yong Fu S. C. Jardin.
Barry Smith Algebraic Solvers in FASTMath Part 1 – Introduction to several packages unable to present today Part 2 – Portable Extensible Toolkit for Scientific.
Parallel coupling: problems arising in the context of magnetic fusion John R. Cary Professor, University of Colorado CEO, Tech-X Corporation.
Presented by Adaptive Hybrid Mesh Refinement for Multiphysics Applications Ahmed Khamayseh and Valmor de Almeida Computer Science and Mathematics Division.
TR&D 2: NUMERICAL TOOLS FOR MODELING IN CELL BIOLOGY Software development: Jim Schaff Fei Gao Frank Morgan Math & Physics: Boris Slepchenko Diana Resasco.
Xolotl: A New Plasma Facing Component Simulator Scott Forest Hull II Jr. Software Developer Oak Ridge National Laboratory
1 1  Capabilities: Serial (thread-safe), shared-memory (SuperLU_MT, OpenMP or Pthreads), distributed-memory (SuperLU_DIST, hybrid MPI+ OpenM + CUDA).
CS 290H Lecture 15 GESP concluded Final presentations for survey projects next Tue and Thu 20-minute talk with at least 5 min for questions and discussion.
Report from LBNL TOPS Meeting TOPS/ – 2Investigators  Staff Members:  Parry Husbands  Sherry Li  Osni Marques  Esmond G. Ng 
Algebraic Solvers in FASTMath Argonne Training Program on Extreme-Scale Computing August 2015.
Consider Preconditioning – Basic Principles Basic Idea: is to use Krylov subspace method (CG, GMRES, MINRES …) on a modified system such as The matrix.
Presented by High Performance Computing in Magnetic Fusion Energy Research Donald B. Batchelor RF Theory Plasma Theory Group Fusion Energy Division.
C OMPUTATIONAL R ESEARCH D IVISION 1 Defining Software Requirements for Scientific Computing Phillip Colella Applied Numerical Algorithms Group Lawrence.
Center for Extended MHD Modeling (PI: S. Jardin, PPPL) –Two extensively developed fully 3-D nonlinear MHD codes, NIMROD and M3D formed the basis for further.
Nonlinear Simulations of Energetic Particle-driven Modes in Tokamaks Guoyong Fu Princeton Plasma Physics Laboratory Princeton, NJ, USA In collaboration.
On the Path to Trinity - Experiences Bringing Codes to the Next Generation ASC Platform Courtenay T. Vaughan and Simon D. Hammond Sandia National Laboratories.
Conjugate gradient iteration One matrix-vector multiplication per iteration Two vector dot products per iteration Four n-vectors of working storage x 0.
TEMPLATE DESIGN © H. Che 2, E. D’Azevedo 1, M. Sekachev 3, K. Wong 3 1 Oak Ridge National Laboratory, 2 Chinese University.
Unstructured Meshing Tools for Fusion Plasma Simulations
Challenges in Electromagnetic Modeling Scalable Solvers
Parallel Unstructured Mesh Infrastructure
Scientific Computing Lab
Meros: Software for Block Preconditioning the Navier-Stokes Equations
Adaptive Grid Generation for Magnetically Confined Plasmas
Integrated Modeling for Burning Plasmas
Presentation transcript:

1 1  Capabilities: Serial (C), shared-memory (OpenMP or Pthreads), distributed-memory (hybrid MPI+ OpenM + CUDA). All have Fortran interface. Sparse LU decomposition, triangular solution with multiple right-hand sides Incomplete LU (ILU) preconditioner (serial SuperLU) Sparsity-preserving ordering  Minimum degree ordering applied to A T A or A T +A [MMD, Liu `85]  ‘Nested-dissection’ applied to A T A or A T +A [(Par)Metis, (PT)-Scotch] User-controllable pivoting: partial pivoting, threshold pivoting, static pivoting Condition number estimation Iterative refinement Componentwise error bounds  Download:  Further information: contact: Sherry Li, Developers: Sherrry Li, Jim Demmel, John Gilbert, Laura Grigori, Piush Sao, Meiyue Shao, Ichitaro Yamazaki SuperLU – supernodal sparse LU linear solver

2 2  A new fusion reactor ITER is being constructed in France, which is used to study how to harness fusion, creating clean energy using nearly inexhaustible hydrogen as the fuel. If successful, ITER will produce 10 times as much energy than it uses — but that success hinges on accurately designing the device.  One major simulation goal is to predict microscopic MHD instabilities of burning plasma in ITER. This involves solving extended and nonlinear Magnetohydrodynamics equations. Two fluid-based codes have been developed, using different finite element discretization methods: M3D-C1 (Steve Jardin et al. PPPL), and NIMROD (Carl Sovinec, Univ. Wisconsin, and others)  These are truly multiphysics simulations involving several coupled PDEs: continuity, Maxwell, momentum, electron energy, ion energy, etc. They are also of multiscale with extremely wide range of time and spatial scales, and high anisotropy. Therefore, the discretized linear systems are indefinite and very ill-conditioned.  SuperLU_DIST has been used in the production codes. For example, the NIMROD simulation time was improved from 3 weeks to 3 days, compared to using some preconditioned iterative solvers. In many iterations of nonlinear solver, the largest linear systems are of dimension 500K, complex and nonsymmetric. SuperLU: Application highlight in fusion energy research  R Z

3 3  There were over 27,000 downloads of SuperLU in FY It was the third most-downloaded software at LBNL Over the years, it has consistently been the most-downloaded software in the Computing Sciences organization at LBNL  It has been used in many high-end simulation codes, such as ASCEM (reactive flow and transport), M3D-C 1 /NIMROD (fusion tokamak design), Omega3P (accelerator cavity design), OpenSees (earthquake engineering), PHOENIX (modeling of stellar and planetary atmospheres).  It is adopted in many commercial mathematical libraries and simulation software, including AMD (circuit simulation), Boeing (aircraft design), Chevron, ExxonMobile (geology), Cray's LibSci, FEMLAB, HP's MathLib, IMSL, NAG, OptimaNumerics, Python (SciPy), Walt Disney Feature Animation. SuperLU: usage and impact